Signal Preprocessing Towards IoT Acoustic Data for Farm Pest Detection

In Africa the realization of the Sustainable Development Goal 2 (SDG-2) of zero hunger by 2030 is threatened by agricultural pests. Acoustic technology offers one method for pest detection in agriculture. AudioMoth microphones were deployed in three farms in Rwanda. Band filters were applied to the...

Full description

Saved in:
Bibliographic Details
Published inIEEE EUROCON 2023 - 20th International Conference on Smart Technologies pp. 78 - 83
Main Authors Amenyedzi, Destiny Kwabla, Kazeneza, Micheline, Vodacek, Anthony, Maginga, Theofrida Julius, Nzanywayingoma, Frederic, Nsengiyumva, Philibert, Bamurigire, Peace, Ndashimye, Emmanuel
Format Conference Proceeding
LanguageEnglish
Published IEEE 06.07.2023
Subjects
Acoustics
Africa
Agriculture
audio signal processing
AudioMoth
Birds
Image edge detection
Labeling
Machine learning
pest
Rwanda
Online AccessGet full text

Cover

More Information
Summary:In Africa the realization of the Sustainable Development Goal 2 (SDG-2) of zero hunger by 2030 is threatened by agricultural pests. Acoustic technology offers one method for pest detection in agriculture. AudioMoth microphones were deployed in three farms in Rwanda. Band filters were applied to the files for listening to pest calls for labeling. The WAV files were processed using Scikit-maad package in Python to spectrograms, which allowed for visualization. The regions of interest were selected to be used for labeling. Results suggest that birds are the dominant pest during the morning sections of the day while frogs are active at night. The contribution of this work is to provide a roadmap towards labeling for detection of agricultural pests using machine learning at the edge.
DOI:10.1109/EUROCON56442.2023.10198956